-
NECESSITY AND TELEOLOGY IN ARISTOTLES PHYSICS
Jacob Rosen
A DISSERTATION
PRESENTED TO THE FACULTY
OF PRINCETON UNIVERSITY
IN CANDIDACY FOR THE DEGREE
OF DOCTOR OF PHILOSOPHY
RECOMMENDED FOR ACCEPTANCE
BY THE DEPARTMENT OF
PHILOSOPHY
Advisers: Hendrik Lorenz and Benjamin Morison
November
-
Copyright by Jacob Rosen,
-
ABSTRACT
Some of Aristotle's arguments for teleology involve a
distinction between twoways of being necessary: it seems that being
necessary in one of these ways pre-cludes being for the sake of
something, while being necessary in the other way en-tails it. is
second way, which consists in being necessary for the achievement
ofan end, is occasionally (six times in all) referred to as a
matter of hypothetical( ) necessity. I inquire into the meaning of
this phrase, beginningwith a survey of the uses to which Aristotle,
throughout his writings, puts the no-tion of a hypothesis. One
upshot is that hypothetically necessary ought simplyto mean
necessary on an assumption, where (nearly enough) q is necessary on
theassumption that p i (i) not necessarily q and (ii) necessarily,
if p then q. e onlypassage where it is really dicult to understand
the phrase this waywhere one istempted to think it needs a richer
meaning, such as necessary for the achievementof an endis the rst
half of Physics II.. I oer a close reading of this passage,one of
whose virtues is that it preserves the phrases broad,
straightforwardmeaning. Finally I consider how widely hypothetical
necessity, thus broadly inter-preted, reaches in the natural world
according to Aristotle. I suggest that there israther less of it
than is generally supposed: this partly explains why it is called
byname only in connection with end-related cases. Regarding the
anti-teleologicalkind of necessitywhich might appear to involve
necessitation of later states byantecedent conditions, and so to be
a species of hypothetical necessityI arguethat it is not something
we would call necessity at all. I work to elucidate what itis, as
well as why and how far it is incompatible with teleological
relations.
iii
-
CONTENTS
Abstract iii
Introduction .. What does hypothetically necessary mean ..
Teleology and necessity .. Background: causation in Aristotle
Chapter . e meaning of hypothetically necessary .. Where and how
oen hypothesis appears .. Hypothesis in the Organon .. Hypothesis
in physical writings and Metaphysics .. Rhetoric: a hypothesis need
not be a proposition .. Politics: can hypothesis mean end? .. From
a hypothesis .. Hypothetically necessary
Chapter . Reading Physics II. .. e opening question .. e wall
theory .. Disagreement .. Disagreement .. e relation between
disagreements and .. Extrapolating general principles ..
Compatibilism? .. Remaining questions of interpretation
Chapter . Backward- and forward-owing necessity .. e meaning of
hypothetical .. Being due to hypothetical necessity .. Necessity in
Parts of Animals .. One-directional necessitation? .. Intuitive
plausibility .. Result
Conclusion
Bibliography
iv
-
INTRODUCTION
One of Aristotles techniques for articulating his teleological
picture of the natural
world, and for clarifying how it diered from other scientic
views of his day, is
by invoking and distinguishing dierent senses of necessity. In
one sense of the
term, Aristotle regards necessity as an exclusive alternative to
teleology: he will
write not for the sake of something, but from necessity (Physics
II., ),
or not from necessity, but for the sake of something (Generation
of Animals II.,
). On the other hand, he sometimes calls something necessaryas
do
weprecisely because it is required for the achievement of an
end. (You can invite
someone to justify an action by asking, was that really
necessary?) In these cas-
es, necessity entails teleology.
ese two ways of being necessary are explicitly contrasted in
Physics II., and
again in the rst chapter of Parts of Animals. It is not clear
quite what the rst sort
of necessity, the anti-teleological kind, is. e second kind is
clearer, in a way, but
it is hard to see how it is supposed to help explain natural
phenomena that do not
involve any deliberation or intentional action. We can
understand how the fact
that stone is necessary for the building of a house could gure
in explaining an or-
der sent o by a house-builder to the local quarry; but how does
the fact, say, that
roots are necessary for nourishment help explain a plants
putting down roots?
e house-builder can do some reasoning: this is to be done, this
requires that,
therefore . A plant does no such thing. us it is a challenge to
learn something
specically about Aristotles natural teleology from his
discussions of necessity.
.
As my starting point, I raise a question not so much about what
Aristotles doc-
trine was, as about the language he uses to express it. He
appears to refer to the
-
end-related kind of necessity, in contrast to the other, as
hypothetical ( -
). is is curious, because on the face of it hypothetically
necessary should
simply mean necessary on an assumption, or conditionally
necessaryand being
conditional is hardly distinctive of end-related necessity. e
other sort, if it is re-
ally necessity at all, should be conditional as well: it is
necessary for this to happen
now if such-and-such conditions obtained earlier. A popular
solution is to main-
tain that hypothetically necessary does not simply mean
conditionally necessary,
but rather something more elaborate: for example, necessary as a
conditio sine
qua non for the achievement of an end. But little argument has
been given, be-
yond appeals to its immediate context, that the phrase has this
more elaborate
meaning; and no one has explained how it could, given the
meanings of its con-
stituent words.
In chapter one, I survey the ways in which Aristotle uses the
word hypothe-
sis () throughout his writings. e description in Bonitz Index
Aris-
totelicus is correct: a hypothesis is something assumed as a
basis for something
(id quod ponitur tamquam fundamentum, Bonitz () ). Oen it is
a
proposition assumed as a premise for some course of reasoning
(sections . and
.), but other sorts of thing may also be assumed, and for other
purposes (section
.). Sometimes what is assumed is an end, or a proposition
stating that some end
is achieved; but hypothesis no more means end in such cases than
father
means carpenter when it refers to a carpenter. To be sure of
this, I look carefully at
the relevant passages (they are in the Politics: statements such
as the hypothesis
of democracy is freedom), and show that they all make good sense
with hypoth-
esis meaning assumption (section .).
If hypothesis never means end, it is hard to see how a phrase of
the form
+ hypothesis could mean anything like for the achievement of
an
end.
. See section ., including note on page , for references.
-
Focusing specically on the prepositional phrase (hypotheti-
cally), I check the passages in which it modies something other
than necessity
(section .). Some are very obscure, but the intelligible
passages are all intelligible
when we take to mean on an assumption. It seems that
hypotheti-
cally necessary means no more than necessary on an assumption.
ere is no
great diculty interpreting it this way when it appears in Parts
of Animals, Gener-
ation and Corruption, and De Somno (section .).
at leaves the rst half of Physics II., one of the passages in
which Aristotle
contrasts the anti-teleological and end-related kinds of
necessity. One task of
chapter two is to resist the appearance that hypothetical
necessity refers speci-
cally to the latter in contrast to the former. is my reading
does: it results that
Aristotles train of thought turns out poorly signposted, but
still cohesive and in
decent order (section .). is leaves us with a choice: either say
that a technical
term is used with a meaning nowhere securely attested (i.e.,
that hypothetically
necessary means necessary for the achievement of an end), or say
that a passage
has misleading signposts. I opt for the latter, but what is most
important is to see
the choice. Aristotles work quite generally displays a rare
mixture of precision and
sloppiness, and as interpreters we need to have a view as to
which are the ways in
which he is careful and which are the ways in which he is
not.
Now if I am right that hypothetically necessary means nothing
more than
necessary on an assumption, then we need to explain why the
phrase is used exclu-
sively to describe things which are necessary relative to an end
(chapter three).
Why does Aristotle never say, for example, that Socrates death
was necessary on
the assumption that he drank hemlock, and therefore
hypothetically necessary?
We could put it down to coincidence: aer all, the phrase
hypothetically neces-
sary appears only six times in Aristotles entire corpus. Or
perhaps Aristotle in-
. Physics II., , ; Parts of Animals ., , ; Generation and
Corruption., ; De Somno , .
-
troduced the phrase during some period of fascination with the
parallel between
natural teleological explanations, on the one hand, and chains
of practical rea-
soning on the other, the latter of which, on Aristotles model,
precisely involve
deducing what is to be done now from the hypothesis of an ends
future achieve-
ment. In contrast there may have been no practice, as there is
in physics class-
rooms today, of assuming initial conditions and deducing the
later state of a sys-
tem. But a third explanation would be that, according to
Aristotles substantive
views about nature, every interesting case of conditional
necessity in nature just is
a case in which it is necessary for something to obtain or occur
if some given end
will be realized. Maybe he didnt believe that earlier conditions
necessitate later
states.
e last proposal may seem like a non-starter, given that
Aristotle oen writes
in his biological works of things coming about from necessity
based upon an-
tecedent conditions, such as a mans head being moist, or an
animal being fright-
ened (section .). But if, as I suggest in section . and argue
more fully in .,
these passages employ a version of (so-called) necessity that is
not really necessity
at all, then the proposal turns out to have some plausibility. I
explore how it can
be supported, and what qualications are needed to make it
defensible (sections
..).
All this work shows that we do well to take the phrase to have
a
consistent meaning across dierent contexts, and a meaning that
arises composi-
tionally from the meanings of its component words. Besides the
immediate conse-
quences to the interpretation of Physics II. and Parts of
Animals I., this result has
broader methodological implications. is not the only
technical
phrase that interpreters have tended to interpret according to
the convenience of
the passage immediately before them, while exerting less eort
than they might to
discover a unied meaning of the term throughout the corpus. Of
course there is
no guarantee that Aristotles linguistic usage is fully
consistent; sometimes
pressures of context must win out over the desire for
uniformity. However, re-
-
sources such as the TLG make it easier than ever before to
examine every appear-
ance of a term, and try to nd out whether one meaning can
account for all of
them. When dealing with a term of art in particular, not one of
natural language, I
think we should be biased toward ascribing the fewest dierent
meanings possi-
ble. is dissertation is an experiment in applying that bias to
the interpretation
of hypothetical necessity. I think it has been fairly
successful, and aim in the fu-
ture to apply similar methods to other terms, such as
(acciden-
tally) and (simply, without qualication).
.
My second object is to clarify Aristotles view of natural
teleology itself, and how
he understands its dierence from others that were current during
his time. I take
somewhat dierent approaches to this question in chapter (on
Physics II.) and
in chapter (on Parts of Animals I.).
.. e intrinsic direction of causes and changes
At the beginning of Physics II., Aristotle illustrates (or
rather parodies) a type of
explanation which purports to account for a things generation
wholly in terms of
an anti-teleological kind of necessity. What exactly are the
essential features of
this type of explanation? I believe they are best understood in
terms of intrinsic
direction toward an end, or lack thereof, on the part of
activities and of the e-
cient causal powers that bring them about. In a nutshell, a
process occurs from
necessity in the present sense just in case every basic process
from which it is
composed is intrinsically aimless. Aristotle seems to presuppose
a plausible prin-
ciple linking processes to their ecient causes, namely that a
basic process is in-
trinsically aimless just in case its proximate ecient cause is
an intrinsically aim-
less power. (To get a sense of the dierence between an
intrinsically aimless power
-
and an end-directed power, compare fragility with the power for
photosynthesis.
Of course fragility may serve an end in the design of some
complex objectper-
haps a fusebut in its own right, I think, it is just a brute
disposition.)
In two other works, Generation of Animals (see ..) and
Generation and Cor-
ruption (see ..), Aristotle makes room for a distinction between
what (if any-
thing) an activity is for intrinsically or in its own right, and
what it may be for in
virtue of the larger context in which it is embedded. In
particular, an intrinsically
aimless activity may count as occurring for an end if it occurs
at the instance or
under the control of a cause or power which is directed towards
that end. (As far
as I can see, this cause will always be a non-proximate cause. e
motions in a
clock seem to me a good illustration of the phenomenon.) is
possibility does
not seem to be countenanced in Physics II.; perhaps this is a
pedagogically moti-
vated simplication, or perhaps the other works were written
later and show a
process of renement in Aristotles ideas.
e key point of dispute between Aristotle and his theoretical
opponents is
presented in this chapter as one about what nature of ecient
causes there are,
and whether or not certain changes and processes are directed
towards the attain-
ment of ends.
.. Order of explanation and of necessitation
Aristotles discussions of hypothetical and other kinds of
necessity have made it
seem to some that the question of teleology hinges crucially on
the temporal di-
rection of necessitation in the natural world: necessitation of
later states by earlier
ones is non-teleological; necessitation of earlier states by
later ones is teleological.
is seems to be a mistake. Suppose there were an infallible
doctor: necessarily, if
he initiates treatment, then the patient recovers. Surely this
does not preclude him
from applying medical treatment for the sake of his patients
health.
-
Instead of necessitation, interpreters may look to order of
explanation. For
example, when Aristotle says that a wall is not due to its
ingredients except as due
to matter, but comes about for the sake of protecting things, he
is oen taken to
be asserting the priority (in some sense) of nal causation over
material causation
(see .).
I have tried to avoid this approach in chapter . I nd these
sorts of claims ob-
scure and not very helpful in getting to the heart of Aristotles
position. It seems
better to move away from ideas about what is more a cause than
what, or what
causes what rst (where the order is not temporal order), and
instead try to ar-
ticulate yes/no questions about whether or not a given cause or
change has a cer-
tain characteristic. Nevertheless, it must be granted that there
is something to the
question of order. is is explicit in Parts of Animals, where
Aristotle raises the
question whether animal generation should be explained in terms
of what the ani-
mal is like once grown, or the other way around. His attitude is
indicated by the
remark,
, - , , .
It is more the case that this and that goes on in house-building
because theform of the house is such-and-such, than that the house
is such-and-such athing because it comes into being in a given way.
(PA I., .)
Given that there is of course causation in both directionsnal
causation
from house to house-building, ecient causation from
house-building to house
it is hard to know what to make of the idea that one is more due
to the other than
vice versa. Still, for all that causal relations come in various
kinds and run in more
than one direction, Aristotles theory of science requires us to
order our knowl-
edge according to a single, one-directional relation of
dependence. In creating a
science of biology, we must settle which facts are to be
demonstrated from which,
. All translations from the Greek are my own, unless otherwise
indicated.
-
and hence must make all-considered judgments as to which facts
are more basic
and more explanatory.
erefore, in chapter I give myself over somewhat to questions of
order. I
propose that Aristotles nature includes no causally interesting
necessitation of lat-
er states by earlier ones, and do my best to make this
proposition plausible. Given
that a science consists of less fundamental truths demonstrated
from more funda-
mental truths, and that demonstration requires necessary
entailment, this has im-
plications as to which propositions will gure as the more
fundamental truths in
an Aristotelian science of biology. In particular, we will not
nd facts about e-
cient causal processes guring as principles from which the
results of those
processes are demonstrated.
. :
Aristotle conceives of being somethings endbeing what it is
foras a way of be-
ing among its causes. So before getting into the main body of
the dissertation, I
would like to make some general remarks about causation in
Aristotle. Some of
what I say will be controversial, but I wont try to argue for it
here. My main pur-
pose in saying it is to articulate a number of presuppositions
that will be in the
background of my discussions later on.
.. Causal relations are metaphysical relations
Because Aristotles notion of an aition or aitia is not exactly
the same as our no-
tion of a cause, scholars have experimented with translations
other than cause. A
popular one is explanation (or explanatory factor, vel sim.). e
motivation is
. Since aitiai are whatever answers a why-question, and whatever
answers a why-question is
-
good, but glossing (what I will continue to call) causation in
terms of explanation
can lead to confusion. Explaining something is an epistemic
activity, whereas cau-
sation is a metaphysical relation out there in the world. We
should be careful both
that we do not identify causal and explanatory relations, and
that we do not rev-
erse their order of priority. Causation is prior to explanation;
the holding of causal
relations is what grounds the correctness of correct
explanations.
Another point is that the appropriateness of an explanation is
subject to very
complicated pragmatic constraints, so that a lot of extraneous,
confusing issues
enter the discussion when we replace questions about causation
with questions
about what is explanatory.
.. e four causes are four causal relations.
Aristotles view is not that there is a single causal relation
and four types of thing
which can stand in this relation to something, or four
(non-causal) relations in
which a cause of X might stand to X. Rather, it is that there
are four dierent
causal relations. Aristotle introduces them as , ways, and oen
refers to
them not with bare nouns such as form or end but with adverbial
phrases such
as as form or as an end (cf. p. , note ). ese are dierent ways
in which
one thing can ground or be responsible for another.
is means that even when the same thing is, for example, the
formal cause of
X and the nal cause of X, there is still a distinction between
formal and nal
cause. e one thing stands in two dierent relations to X: the
relation of formally
causing on the one hand, and the relation of nally causing on
the other. e same
goes when a single thing is the material cause of X and the
moving cause of X.
an explanation, it follows that an aitia is simply an
explanatory factor, whatever this may be,Moravcsik (), p. . It is a
great improvement to cease thinking of an aitia as a cause and
totreat it instead as an explanation, a because, Annas (), p. .
Nussbaum () translatesreason. Barnes () uses both reason and
explanation, explaining himself on pp. .Freeland () and Gotthelf ()
advocate returning to the traditional translation cause.
-
(Hence, if a theorist says that the basic material constituents
of the world are re-
sponsible for all the changes in the world, it would be wrong to
describe that theo-
rist as appealing only to material causation. e theorist appeals
both to material
and to moving causation: its just that he says that the material
and moving causes
in nature are the same stus.)
.. What sorts of things are caused
Aristotles causal talk is quite irregular. However, we can class
many of his causal
ascriptions under four main heads, according to whether what is
being caused is
substantial or non-substantial, and whether it is an
object/feature or a change.
us a cause may be a cause of a ; it may be a cause of a
things
or ; it may be a cause of a things having some feature, in which
case it
is a cause of to ; or it may be a cause of a
non-substantialchange, in which case it will be a cause of F to
.
.. Examples
... Moving causes
: the house-builder is a cause of the house.
. Philoponus in Ph may be making this mistake when he writes
(.-), , , ; If forms invari-ably followed upon the power of the
matter, what need would there be for the productive causeImean,
nature? (emphasis added). Charles (), p. seems to make a similar
slip when hewrites, Many of [Aristotles] predecessors erred, in his
view, in thinking that such phenomena[viz., physical, biological
and psychological] could be explained in terms of material
causationalone. It is true that according to Aristotle most of the
earliest philosophers dealt only with mater-ial causation, i.e.,
with the question what things are made from (Meta. ., ). But in
thepassage Charles quotes, , Aristotle is pointing up the
insuciency of appealing to mater-ial elements as the only moving
causes. Charles himself more correctly describes Aristotles target
afew sentences later as the wish to employ only the resources of
material causation and ecientcausation (involving the matter alone)
(p. , emphasis added).
-
: the house-builder is a cause of the houses coming into
being.
: the house-painter is a cause of being blue to the house.
: the house-painter is a cause of becoming blue to the
house.
... Final causes
: shelter is a cause of the house.
: the house, and therefore shelter, are causes of the houses
com-
ing into being.
: prettiness is a cause of being blue to the house.
: being blue and prettiness are causes of becoming blue to the
house.
... Formal causes
: the capacity to shelter bodies and goods is the formal cause
of the
house.
: the form of house (i.e., the capacity to shelter bodies and
goods) is
the formal cause of the houses coming into being.
: the color blue is the formal cause of being blue to the
house.
: the color blue is the formal cause of becoming blue to the
house.
... Material causes
: wood and stone are causes of the house.
. Assuming a change is goal-directed, the endpoint of the change
will be its immediate nalcause. Whatever the endpoint in turn is
for, will be a remote nal cause of the change. See .... Metaphysics
., . I dont insist that this is a perfectly adequate denition of
house.. Im not sure about this, but my best guess is that the
formal cause of a change is the form ofthe changes endpoint.
(Perhaps if the change is from form to privation, its formal cause
is theform of its starting point.) Cf. Phys II., -.
-
: wood and stone are causes of the houses coming into being
(by
underlying the change in question).
: the house is a cause of being blue to the house (since it is
the sub-
ject, ). Stone is a cause of solidity to the house.
: the house is a cause of becoming blue to the house (since it
is the
subject, , of the change).
. Aristotles examples of material causes in Phys II. are always
the matter of a thing, not thesubject of an attribute. However, his
characterization of the material cause at as that out ofwhich a
thing comes to be and which persists ( ) harkens backto the
discussion of change in Phys I., where the persisting () subject of
an accidentalchange and of its resultant attribute was treated in
the same way as the matter of generation and ofits resultant
substance. Cf. Metaphysics ., -.
-
CHAPTER THE MEANING OF HYPOTHETICALLY NECESSARY
ere are two broad questions in this dissertation. One is about
the meaning of
the phrase hypothetically necessary; the other is about the
nature of Aristotles
natural teleology and his understanding of how it diered from
other current
theoretical approaches. e present chapter is devoted entirely to
the rst. Its job
is to convince us that we should go into the interpretation of
the central text to be
treated in chapter two, namely Physics II., with a strong
presumption as to the
meaning of hypothetically necessary in that text. e presumption
is that this
phrase means simply necessary on an assumption, or conditionally
necessary, and
does not carry any teleological content.
My method is simple. I begin with an overview of Aristotles use
of the word
hypothesis on its own, and show that it never means anything
like end. e only
context in which it has been thought to have such a meaning is
in the Politics,
where Aristotle occasionally refers to the ends of political
constitutions and their
legislators as hypotheses of those constitutions and
legislators. e relevant pas-
sages can be understood perfectly well by taking hypothesis to
have its usual
meaning of assumption or premise. Next, I go through every
appearance of the
prepositional phrase from a hypothesis ( , also translated
hypo-
thetically) in which it modies something other than necessity.
As we should ex-
pect given the meaning of hypothesis on its own, the
prepositional phrase never
means anything like relative to an end. Finally, I discuss the
passages outside of
Physics II. in which Aristotle uses the phrase hypothetically
necessary (-
or , literally necessary from a hy-
pothesis). Although the things called hypothetically necessary
are in fact neces-
sary for the achievement of some end, there is no good reason to
incorporate this
. Bonitz () . Cf. Newman () ad .
-
fact into the meaning of the phrase hypothetically necessary. e
passages are
naturally understood when we take it to mean necessary on an
assumption.
.
e word hypothesis is used times in Aristotles corpus. It occurs
most fre-
quently by far in the Prior Analytics ( times), especially in
discussions of proof
by reductio (where the proposition that is or should be assumed
for reductio is
called a hypothesis) and of hypothetical syllogisms more
generally (these syllo-
gisms or their conclusions are completed or deduced from or
through a hypothe-
sis). e Politics comes in a distant second with occurrences, and
the Posterior
Analytics comes third with .
Of the words appearances, are in the phrase , which I will
translate hypothetically or from a hypothesis. is phrase is used
times in
the Prior Analytics, always describing hypothetical syllogisms,
and it is used in the
same way times in the Topics and probably times in the Posterior
Analytics
(this is in AnPst ., where Aristotle considers whether one could
demonstrate a
denition hypothetically: it isnt completely clear what he has in
mind, but it looks
as though hes talking about hypothetical syllogisms). at leaves
further uses
of , and these are quite varied: Aristotle speaks of knowing
hypo-
thetically, of being good hypothetically, of being false
hypothetically, of being a
citizen hypothetically, and so on. Six times, he speaks of being
necessary hypothet-
. e numbers in this section derive from TLG searches. e TLG
database does not includealternate manuscript readings, and
contains occasional misprints; but for the purposes of a
generaloverview, I think we may have adequate condence in the
numbers. Of course, the frequency withwhich a term appears in a
work is not a sure guide to the importance of the term in that
work. Iam using these numbers merely to give a preliminary
orientation.. From a hypothesis is used to modify four dierent
verbs in such contexts. () show(): AnPr ., , , , . () argue ():
AnPr ., . ()complete (): AnPr ., , ; ., , . () agree (): AnPr
.,.
-
ically. It is this usage that I ultimately want to understand:
what does Aristotle
mean when he calls something hypothetically necessary?
Here is a list of works containing the word hypothesis, along
with the number
of times hypothesis appears in the work.
AnPr EE PA
Pol Frag GA
AnPst Phys lineis
Meta NE Poet
Top Rhet somno
DC GC spiritu
.
e paradigmatic uses of hypothesis seem to be found in the
logical works.
.. Hypothesis as a kind of deductive principle
e only explicit discussions of what a hypothesis is are in
Posterior Analytics
. and .. ere are some problems of interpretation, but it seems
pretty cer-
tain that the same notion of hypothesis is under discussion in
both passages, and
that the following things are true of it:
. A hypothesis is a proposition ().
. It is a deductive principle ( ), which means or entails
that
a. ere is no demonstration of it, and
-
b. It gures as a premise in demonstrations.
. It gures as a deductive principle for a single special science
(unlike axioms,
which are principles of all sciences, or at least of more than
one science).
A demonstration is a deduction that yields knowledge of its
conclusion. One
condition on this is that its premises be more basic (more
knowable by nature)
than its conclusion. So point () entails that hypotheses are
true (since only
truths can be known) and that they cannot be deduced from any
truths more ba-
sic than they are. However, they are not the most basic truths
there are: a hypothe-
sis is in some sense less fundamental and less self-evident than
an axiom. In .
Aristotle says that that which through itself necessarily is and
necessarily seems
to be is not a hypothesis. Probably he thinks that such a thing
is an axiom.
us an axiom, such as the proposition that one thing cannot both
belong and not
belong to another thing at the same time and in the same
respect, would be true
through itself and would have to be seen to be true through
itself. In contrast, a
hypothesis fails at least one of these two conditions. Perhaps
one acquires con-
dence in a hypothesis by induction () rather than by mere
consideration
of the proposition itself, so that it is not through itself that
the hypothesis seems to
be the case.
Does a hypothesis also fail to be through itself? It is not
clear what this would
amount to. Aristotle could hold that hypotheses are true in
virtue of some other
. See Barnes () ad (pp. ) for the question whether axioms must
be common toall sciences or just more than one.. AnPst ., . AnPst
., ; cf. -.. A., -. .. Barnes (), p. thinks that Aristotle is
referring to all the principles of a science, not justaxioms, and
that hypothesis in AnPst A. does not mean what it meant in A., i.e.
a kind ofprinciple of a science. is seems to me to rest on a
misunderstanding of , for which seethe next section and note .. But
see Bolton (), p. for an argument that propositions known
inductively are knownthrough themselves (emphasis original).
-
truth(s) obtaining, but it is not clear what those other truths
would be. ey must
not entail the hypothesis, since otherwise it would not be a
deductive principle (it
could be deduced from truths more basic than it, and that would
be to demon-
strate it). Could they be the individual instances of the
general truth stated by the
hypothesis? (A holds of this B, A holds of that B, etc., where
the hypothesis is
A holds of all B?) However exactly it works, a hypothesis is a
second rate princi-
ple, a runner-up to axioms.
In addition to distinguishing them from axioms, Aristotle also
distinguishes
hypotheses from denitions (). A denition, like a hypothesis and
unlike
an axiom, is a principle of a single science. Aristotles
descriptions of the dier-
ence between hypotheses and denitions are hard to sort out, and
it would be dis-
tracting to worry about them now. In some sense or other, a
denition is sup-
posed to make clear what something signies, whereas a hypothesis
says that
something is or is not. It is unclear whether this means that
denitions are not
propositions whereas hypotheses are, or whether denitions and
hypotheses are
two dierent kinds of proposition (e.g., with denitions stating
that such-and-
such is the essence of a kind, and hypotheses stating that a
kind exists or that a
given item is a member of a given kind).
We can now add two more features of hypotheses as treated in
Posterior Ana-
lytics . and . Since these two pretty much complete the picture,
I will re-
produce the full list.
. A hypothesis is a proposition.
.a.It says that something is or is not (unlike a denition).
. It is a deductive principle, which means or entails:
.a. ere is no demonstration of it;
.b. It gures as a premise in demonstrations.
. See emistius in AnPst .; Philoponus in AnPst .; Barnes (), pp.
-.
-
. It gures as a deductive principle for a single special science
(unlike an axiom).
. Its truth is not maximally self-evident or self-grounding
(unlike an axiom).
... Relative hypotheses
At a certain point in ., Aristotle considers a case in which a
demonstrable
proposition is not demonstrated, but simply assumed. He has a
didactic context
in mind: a teacher might invite a pupil simply to accept
something as true without
proof (presumably, justication is to be given later). When this
happens, Aristotle
says, the proposition assumed is not a hypothesis full stop, but
it is a hypothesis
relative to the student ().
is seems exactly the right thing to say, given the
characterization of hy-
potheses I have been summarizing. Speaking without qualication,
there is a
demonstration of the proposition, and so it is not a deductive
principle (see .a
above). Relative to the learner, on the other hand, there is no
demonstration of it.
e learner deduces things from it, but does not deduce it from
other things; it
functions for him as a principle.
.. Reduction to the impossible and other hypothetical
syllogisms
Although the usage Ive been discussing is the one that is best
described in
Aristotle, it is not the one that is most commonly used. e most
numerous use of
hypothesis is in the Prior Analytics, and this in two contexts.
One is the discus-
. , .. Barnes, by contrast, seems to take Aristotle to be
introducing a new sense of hypothesis(the type of supposition dened
here, Barnes (), p. ), on which it is a relative term. ismisses the
force of lines , which describe the terms relative application
precisely in termsof its primary, unqualied use: it is a hypothesis
not without qualication, but only in relation tothat person. Waitz
(), p. too misses the contrast between and , so thatin his eyes
Aristotles statements in . and . inter se videntur pugnare.
-
sion of so-called hypothetical syllogisms; the other is the
discussion of proof by
reduction to the impossible. In neither context is it plausible
that Aristotles hy-
potheses are intended to be deductive principles.
... Hypothetical syllogisms
Hypothetical syllogisms are ones in which the premises somehow
do not en-
tail the conclusion in a fully direct way; instead, the
conclusion is drawn or
proven from or through some additional hypothesis ( , -
). e primary examples are arguments by reductio, which seem to
rely on
the hypothesis that whatever entails something false (or
impossible) is itself false
(or impossible). In the Topics Aristotle mentions arguments in
which a conclusion
is rst drawn for some subset of the items of interest, and then
generalized based
on the hypothesis that whatever holds of some holds of all.
Aristotle doesnt spell out in detail what the role of the
hypothesis is in a hypo-
thetical syllogism. My impression is that it does not gure as a
premise, but rather
serves to license an inferential move. If so, then it is clearly
not functioning as a
hypothesis of the kind described in Posterior Analytics A. and ,
since that kind
of hypothesis was said to serve as a premise in demonstrations.
Furthermore, ar-
guments by reductio can be employed in the examination of any
subject matter, so
that the hypothesis they depend on must not be proper to any one
special science.
. Topics ., . Smith (), p. thinks the hypothesis does serve as a
premise: in his comment on A., he writes, us, these arguments are
deductive only from (ek) an assumption, i.e.,deduce from an
assumption as a premise: they are not really deductions of their
ultimate conclu-sions, but of something else. In Ross (), pp. -,
hypothetical argument is analyzed into asyllogistic and a
non-syllogistic part, with the hypothesis being used in the
non-syllogistic part,hence not at least as a formal premise.
-
... Hypothesis as what is assumed for reductio
One very common context for hypothesis is when Aristotle is
discussing
proof by reduction to the impossible. He uses the word to refer
to the proposition
that is or should be assumed for reductio, i.e., the
contradictory of the proposition
to be proved, or occasionally to a proposition that might
mistakenly be assumed,
such as the contrary of the proposition to be proved. Again,
these are obviously
not hypotheses of the kind described in Posterior Analytics A.
and , since those
are true and are principles, whereas these are false (or at
least taken by the arguer
to be false), and are assumed in order then to be rejected, not
as premises from
which to infer further truths.
.
In the physical writings and in the Metaphysics, the meaning of
hypothesis typi-
cally seems close to the one given in Posterior Analytics A. and
. A hypothesis
is a proposition taken to be true without proof, and used as a
more or less funda-
mental premise.
In some contexts it looks as though the hypothesis might count
among the
principles of a special science. For example, in De Caelo,
Aristotle works with an
assumption as to what simple motions there are, and the
assumption that for each
simple body, a dierent simple motion is natural to it. He refers
to these as the
rst or primary () hypotheses, and as the hypotheses concerning
the
motions. In the Physics, he says that a special scientist need
not concern himself
. Contradictory: AnPr ., , ; ., , ; ., passim; ., passim.
Contrary:AnPr ., .. e only simple motions are straight and
circular, De Caelo ., . ere is a singlenatural motion of each of
the simple bodies, De Caelo ., , .. De Caelo ., and .. De Caelo .,
.
-
with objections to the principles of his science, and that in
physics it is a hypoth-
esis that nature is a principle of change. In Metaphysics ., the
Philosophical
Lexicon entry on principle (), the hypotheses of demonstrations
are cited
as examples of principles in the sense of that from which a
thing is knowable pri-
marily. In these passages a hypothesis is at least a quite
fundamental premise,
and possibly a deductive principle of a science.
In Metaphysics and , Aristotle speaks several times of the
hypotheses
maintained by believers in Forms and/or separate mathematical
objects. Since
the hypotheses are false (., ), they cannot be genuine deductive
princi-
ples; but they seem to be fundamental premises of theories, at
any rate.
In places it isnt clear how fundamental an assumption a
hypothesis is meant
to be, where it lies on the continuum between a principle of a
science and a merely
local, tentative assumption. Examples: () It is clear that the
surface of water is
like this [sc. spherical] if we take as a hypothesis that water
by nature always ows
into what is more hollow: and what is nearer to the center is
more hollow. ()
Certain theorists say things contrary to mathematical doctrines
(
); and yet it is right () either not to displace them [sc. the
doc-
trines], or to do so by means of accounts more trustworthy than
the hypothe-
ses. () In Generation of Animals, Aristotle proposes three
hypotheses about the
generative motions imparted by parents, in order to explain why
ospring may re-
semble one or another ancestor to a greater or lesser degree. It
is not clear
whether these are meant to be deep principles, or just plausible
ad hoc assump-
. Physics VIII., -. Metaphysics ., -. , , .. Metaphysics M, , ;
; , . N, , .. De Caelo ., -.. De Caelo ., -. .. GA IV., -.
-
tions which Aristotle hoped eventually to replace or derive from
more basic
truths.
. :
According to Bonitz, in rhetoric the word hypothesis signies the
topic under
discussion (eam rem signicat de qua agitur, -). us at Rhetoric
III.,
, Aristotle says that unusual and high-sounding words are more
appropri-
ate to poetry than to prose, because in prose the hypothesis is
lesser than in poetry.
Roberts (in the Oxford translation edited by Barnes) oers
subject-matter as a
translation of hypothesis here.
Bonitz also cites the phrase speak in relation to a hypothesis
(
), at Rhet II., , as an example of this usage. In fact, it isnt
clear that
the word has the same meaning here. Aristotle has just said that
the same meth-
ods are applicable whether one must convince many judges or only
one; he now
says, and likewise whether one is addressing someone who
disagrees or a hypoth-
esis. Hypothesis does not seem to mean the topic under
discussion, as Bonitz
suggests. Perhaps it refers to an assumed opponent, as opposed
to an actual one.
In any case, these examples indicate that a hypothesis is not
always a proposi-
tion. Linguistically, the word should signify an act of laying
down or positing or
assuming (), and by extension the result of such an act, i.e.,
something
laid down or posited or assumed. We know that propositions are
not the only
thing that one can assume. For instance, Aristotle says in NE
VI. that cleverness
is the ability to achieve any assumed goal ( , ).
. , . -.. is is parallel to how means the activity of growing
and also that which grows (PhysicsII., ) and how means both the
activity of making and what is made (specicallypoetry).
-
. :
Bonitz claims that in Aristotles political doctrine, the notion
of a hypothesis is
close to the notions of an end or goal () on the one hand, and a
dening
characteristic () on the other. (e notions of end and dening
characteris-
tic are, of course, not equivalent. e idea must be that any kind
of constitution
will have some feature that it both aims to instantiate (the
feature is an end), and
to some degree succeeds in instantiating (the feature is a
dening characteristic):
for example, a constitution is a democracy only if it aims at,
and to some extent
succeeds in achieving, freedom.) is claim should be considered
carefully. For if
there are any contexts in which hypothesis really means
something like end, this
would be evidence that necessary from a hypothesis can mean
something along
the lines of necessary for an end.
However, Bonitz himself explains Aristotles usage in a way that
makes clear
how the word hypothesis can be used to refer to ends without
meaning end.
e explanation, to quote Bonitz quoting Aristotle, is that in
actions, that for the
sake of which is a principle, just as hypotheses are in
mathematics. is state-
ment is based in Aristotles model of practical (and productive)
reasoning, accord-
ing to which the reasoner begins from a premise describing the
realization of an
end, and makes a series of inferences (presumably relying on
additional premis-
. in doctrina politica (quoniam , . ) non multum diert a
notionibus et.. Of course, means from, not for; but the phrase
could be taken to express the thought thatsomething is necessitated
by, its necessity derives from, an end.. Bonitz , quoting NE ., ..
e premise in reasoning seems to be that an end must or needs to be
realized. In Movementof Animals ch. , Aristotles examples involve
verbal adjectives ( , ; , ; , ; ,-; , ) or the verb need ( , -). NE
. gives anexample using a premise about what is advantageous ( ,
-)and an example involving a premise about what is required ( ,).
On the other hand, when something is necessary on a hypothesis, the
hypothesis would
-
es about what means are available and eective) until she arrives
at the specica-
tion of something it is now in her power to do. Given this
model, when one
speaks of premises or assumptions in connection with practical
matters, it will be
clear that one is referring either to propositions about ends or
to propositions
about the available means to those ends. Oen, it will be clear
that one is referring
specically to propositions about ends. For example, if I speak
of the hypothesis
of a democratic legislator in contrast to that of a tyrant, it
will be clear that I am
talking about his end, since the interesting dierences between a
democrat and a
tyrant lie in their ends, not in the means available to
them.
Here, then, is a possible explanation for Aristotles use of
hypothesis in the
Politics to refer to the ends of political actors and regimes.
Let us look at the pas-
sages and see whether the explanation succeeds.
.. Socrates hypothesis in the Republic
Perhaps the best place to start is with Aristotles criticism of
Platos Republic,
since the role of a political hypothesis as a starting point for
argument is easiest to
see there. In Politics II., Aristotle raises a number of
objections to the sharing
of wives, children, and property advocated in Republic book .
His objections are
of two broad types. On the one hand, he rejects the basis on
which Platos Socrates
rests his arguments for communism, namely the claim that a city
should be as
unied as possible. On the other hand, he argues that Socrates
communist
arrangements would not conduce to unity in any case. We can
think of the rst
sort of objection as external, and the second as internal (since
it aims to show that
Socrates proposals are unsuccessful by his own standards). Here
is how Aristotle
introduces his discussion:
seem to be the proposition that an end is or will be achieved,
not the proposition that it should ormust be achieved. (It is not
necessary that if there should be a house tomorrow, then
foundationsare laid today.)
-
, , . , , , , , -. - .
ere are many diculties with having everybodys wives be common,
and inparticular, the reason for which Socrates says that things
should be legislatedthis way can be seen not to result from his
logoi (words, arguments, discus-sion). Moreover, the end which he
says the city must attain is impossible ashe states it, and how it
should be interpreted is not at all determined. I mean,for the
entire city to be as far as possible one, on the grounds that this
is best:Socrates takes this as his hypothesis. (Politics II., )
e rst sentence announces an internal criticism: Socrates reason
for his leg-
islation, namely unity for the city, will not be achieved
through the measures he
describes (cf. Pol II.). e second sentence announces an external
criticism: it is
impossible for a city to be fully one, since, as Aristotle will
go on to say, whatever
surpasses a certain degree of unity is ipso facto not a city.
Hence perfect unity is
not a correct aim for legislation.
Now Aristotle refers to the citys unity as Socrates hypothesis;
and he does so
again in chapter II.:
. I take logoi to refer ambiguously to Socrates descriptions of
communist measures, and his ar-gument () that those measures will
lead to unity by arranging for all citizens to be pleasedand pained
by the same things.. Jowett (Barnes (), p. ) and Newman () vol , p.
understand the two sen-tences the other way around, with the rst
indicating that unity is not shown by Socrates to be de-sirable,
and the second that Socrates means towards that end are impossible
to implement. Jowetttranslates, the principle on which Socrates
rests the necessity of such an institution evidently isnot
established by his arguments. Further, as a means to the end which
he ascribes to the state, thescheme, taken literally, is
impracticable, and how we are to interpret it is nowhere precisely
stated.Against this, there are no arguments for the goodness of
civic unity in the Republic, and Socrates infact refers to its
goodness as a starting point for discussion ( , Rep ). Also,the
language of the rst sentence is mirrored at the beginning of
chapter , where Aristotle giveshis internal criticism.Reeve ()
altogether fails to see that two types of criticism are being
announced, translating itis not evident from Socrates arguments why
he thinks this legislation is needed. Besides, the endhe says his
city-state should have is impossible.
-
- . , .
One must think that the cause of Socrates error is that his
hypothesis is notcorrect. Both a household and a city must be one
in some way, but not in everyway. (Pol II., )
It is easy to see why the label hypothesis is appropriate if we
look at the passage
in the Republic to which Aristotle is most likely referring. is
is the passage in
which Socrates begins his defense of communism.
, - , , , () , ; , . (b.) ; - ; .
en isnt the rst step towards agreement to ask ourselves what we
say is thegreatest good in designing the citythe good at which the
legislator aims inmaking the lawsand what is the greatest evil? And
isnt the next step toexamine whether the system weve just described
ts into the tracks of thegood and not into those of the bad?
Absolutely. Is there any greater evilwe can mention for a city than
that which tears it apart and makes it many in-stead of one? Or any
greater good than that which binds it together and makesit one? ere
isnt. (Rep V, , translation by Grube/Reeve in Cooper())
Socrates and his interlocutor Glaucon agree to begin their
discussion of the
merits of communism from the premise that the greatest good for
a city is what-
ever unies it, and the greatest evil for a city is whatever
destroys its unity. Socra-
tes can then go on to argue that sharing wives, children, and
property is best for a
city by arguing that it unies a city more than any alternative
arrangement. e
goodness of unity thus functions as an unargued premise for
argument, which is
precisely what a hypothesis is.
Aristotles phrasing leaves it ambiguous exactly what he is
labeling a hypothe-
sis. It could be the proposition that it is best for the entire
city to be as far as possi-
-
ble one, or it could be the property of being as far as possible
one. As were about
to see, he sometimes does clearly refer to a property as a
hypothesis.
.. e hypothesis of democracy is freedom
e passage that Bonitz lists rst () as an example of
hypothesis
referring to an end, is Aristotles statement at the beginning of
Politics VI. that
the hypothesis of democracy is freedom:
( , ).
e hypothesis of democracy is freedom (for people are accustomed
to saythis, that only in this constitution do people share in
freedomfor, they say,this is what every democracy aims at).
(Politics VI., -.)
Unlike in the criticism of Plato, here the hypothesis is
ascribed to the constitu-
tion rather than to the person who designs or legislates the
constitution. It is
probably best to understand this ascription as derivative: it is
the hypothesis of the
constitution insofar as the constitution was legislated, or is
structured as if it had
been legislated, by someone who took freedom as a starting point
for his design.
e two things Aristotle says in explanation of his statement that
freedom is
the hypothesis of democracy are, rst, that freedom is a
distinctive feature of
democracy (or at least, people think it is), and second, that
every democracy aims
at freedom (again, at least people think it does). us we can see
why Bonitz takes
this passage to indicate that hypothesis means something akin to
(dening
characteristic) and (end). But again, there is no need to
ascribe a new
meaning to hypothesis: Aristotles line of thought makes good
sense if we take
him to be saying that freedom is the premise of democratic
design and legislation.
Not that every premise behind a constitutions design must
involve something
unique to that constitution: presumably, things such as
stability and auence are
among the basic aims of more than one kind of legislator. But
the ones worth
-
mentioning will be at least distinctive, if not unique (in fact,
Aristotle probably
does not endorse the claim that freedom is unique to democracy).
Second, the
structure of the reasoning involved in designing a constitution
is such that the
starting points are pretty much guaranteed to function as aims.
When the legisla-
tor begins by xing certain features of his constitution, and
then designs the rest
of the constitution in accordance with them, he is in eect
choosing the remain-
ing features (or at least some of them) for the sake of the ones
he xed at the be-
ginning (see pp ).
e hypothesis is said to be a property, freedom, rather than a
proposition. We
could take this as a kind of periphrasis, with the hypothesis
really being a proposi-
tion such as that it is best for the city to be as far as
possible free, or that the city
must be free. But it isnt necessary to do that, since, as weve
seen (section .),
Aristotle allows for non-propositional hypotheses.
.. e tyrants hypotheses
e second passage cited by Bonitz (-) regards the purposes of
tyrants.
, - , , , .
. is is hard to demonstrate, because it is hard to know what
freedom () means inthe present context: is it being used in the
same sense as when Aristotle uses it to signify non-slaves in
contrast to slaves (e.g., Politics I., ); or as when he refers to
the free in contrast tothe wealthy (e.g., Pol IV., : a constitution
is a people [i.e., a democracy] when the freeare in charge, and an
oligarchy when the wealthy are; but it turns out that the former
are many andthe latter few.)? If it is, then Aristotle clearly
thinks that there are many free people under non-de-mocratic
constitutions. In any case, we should note that Aristotle thinks
the typical democraticunderstanding of freedom, as doing whatever
one wishes, is mistaken (Pol V., -). ismakes it probable that he
thinks freedom properly understood is most fully enjoyed under
somenon-democratic type of regime.
-
e purposes of tyrants all refer back to these three terms. For
all tyrannicalelements may be referred to these hypotheses: that
people not trust each other,that they have no power, and that they
think small. (Pol V., -)
As in the remark about democracy, we nd talk of hypotheses
closely associated
with mention of and of purposes (, the three clauses). But
again, the association of these elements does not entail
sameness of meaning. As
for the three , I am inclined to take them as terms, in the
Analytics sense of
components of propositions. Aristotle has in fact given us three
terms in the pre-
ceding lines or so: labels for the characteristics that tyrants
aim to instill in
their subjects. ese are smallness of soul, mistrust of one
another, and powerless-
ness. Aristotle can now conrm that the purposes of tyrants all
refer back to one
or another of these three terms by claiming that every
tyrannical element can be
referred to one or another hypothesis built from them.
Exactly what the three hypotheses are, and just how they are
related to the
clauses, is dicult to determine. ere seem to be two possible
construals of
the syntax of the clauses: they may either be nal clauses or
object clauses
(Goodwin (), ; see for subjunctive in object clauses).
Object
clauses usually follow verbs that signify striving, planning, or
bringing something
about, and can stand in apposition to an object accusative such
as
(Goodwin (), p. ). Final clauses may express the end or purpose
of the ac-
tion of any verb; they would stand in apposition to a phrase
such as ,
for the sake of this, rather than a bare , this.
I think it is better to take the clauses as object clauses, and
ll out Aristotles
somewhat elliptic sentence by understanding in front of each
:
, [ ,] , [ ,] -, [ ,] .
. , ; , ; , .
-
For all tyrannical elements may be referred to these hypotheses:
some to this that people not trust each other, some to this that
they have no power, andsome to this that they think small.
us each (this) has (hypotheses) as its antecedent, and so
each object clause, being in apposition to a , expresses the
content of one of
the three hypotheses. I prefer this interpretation largely
because it feels better to
let , these hypotheses, refer cataphorically ahead to the
subsequent three phrases, rather than anaphorically back to the
mention of three
.
e passage is not straightforward, but in any event, whatever
exactly the hy-
potheses are, it seems unproblematic to maintain that Aristotle
refers to them as
hypotheses because they are laid down as starting points for the
tyrants thought
and action, and not because they are goals or specications of
goals.
Later in the chapter, Aristotle describes an alternative way to
ensure the
longevity of a tyranny, almost opposite to the method we have
just seen. Again
he uses the term hypothesis, but I think in a signicantly
dierent way.
-, , , , . .
. If instead we take the clauses as nal clauses, then I think it
is best to construe and as standing in apposition to , and to
supply a participle such as. us:
, [] , , .
For all tyrannical elements may be referred to these hypotheses,
some havingbeen done in order that people not trust each other,
some in order that they haveno power, some in order that they think
small. (Pol V., a-)
On this construal, we are not told explicitly what the
hypotheses are. Our choices seem to be eitherto supply our own (for
example: they must not trust each other, they must have no power,
theymust think small), or to suppose that is simply picking up from
a line before,and that Aristotle is referring to exactly the same
things (in my view, the terms smallness of soul,mistrust, and
powerlessness).
-
, .
Just as one way of destroying a kingship is to make its rule
more tyrannical, soa way to preserve tyranny is by making it more
king-like, guarding just onething: power, i.e., that he rule not
only those who are willing but also thosewho are not willing (for
if this goes, being a tyrant goes). But while this, like
ahypothesis, must remain, he must either do or seem to do other
things, play-ing well the part of a king. (Politics V., .)
is time the hypothesis (or, that which must remain like a
hypothesis) is not
presented as a premise or origin of everything else the tyrant
does; to the contrary,
it appears to serve as a check and limit on the tyrants actions,
which in themselves
tend in an opposed direction. Indeed, even in its relation to
the constitutional
form of tyranny, the hypothesisthe power to rule both willing
and unwilling
subjectslooks more like a conclusion than a premise. Aristotle
says, if this goes,
being a tyrant goes, which evokes the rule that if a conclusion
doesnt hold then
the premise doesnt hold: being a tyrant ( ) stands in the place
of
premise, and power stands in the place of conclusion.
I would suggest that Aristotle is invoking a dierent context for
the use of hy-
pothesis than those in which hypotheses function as premises.
Consider instead
his descriptions of dialectical confrontations in which the
respondent is said to
uphold (), and the questioner to attack (), a hypothesis.
Aristo-
tles tyrant is doing something like playing the role of
respondent. In taking up the
king-like behaviors and pretenses that Aristotle recommends, he
is not positively
arguing or working for power; rather, he is making what
concessions are required
while doing his utmost to avoid compromising his power.
Aristotle says that the
. e.g., Physics II., -: .. Topics ., -: .e same hypotheses are
dicult to attack as are easy to uphold. Topics ., : . One should
take care not to uphold an implausible hypothesis.
-
tyrant must guard, , this power; this too is a word that comes
up in di-
alectical contexts.
.. Criticizing the hypothesis of Spartas legislator
, - , . - , - .
One could also criticize the hypothesis of [Spartas] lawgiver in
the followingway (this is precisely what Plato has criticized in
the Laws): for the entire sys-tem of laws is oriented towards a
part of virtue, namely military virtue, since itis useful for
conquering. Well then, they were preserved while they were wag-ing
war, but began to perish once they were in power, because they
didntknow how to be at leisure, not having practiced any other kind
of trainingmore elevated than that for war. (Politics II., .)
Based on the corresponding passages in Platos Laws, it appears
that the hypothe-
sis Aristotle has in mind is something to the eect that a city
is well-governed if
and only if it is well-prepared to be victorious in war ( ,
,
, ). It is a consequence of making this his premise that the
law-
giver arranges his laws so as to promote military virtue to the
neglect of other
virtues. e hypothesis is about the importance of military
conquest, and the crit-
icism of that hypothesis is that it results in cultivating a
mere part of human
virtue.
is passage from Laws I indicates the nature of the
hypothesis:
{.} - , , . ;
{.} .
. NE ., : , . Noone would call someone who lived like this
happy, unless he were guarding a thesis.
-
{.} , , ;
A: But explain this point to me rather more precisely: the
denitionyou gave of a well-run state seems to me to demand that its
organization andadministration should be such as to ensure victory
in war over other states.Correct?
C: Of course, and I think our companion supports my
denition.
M: My dear sir, what other answer could one possibly make, if
one is aSpartan? (Plato, Laws , translation by Saunders in Cooper
().)
is passage from Laws III shows that the over-narrow focus on the
military
part of virtue is seen as a consequence of the hypothesis. (See
also Laws IV,
.)
{.} , , - , - .
A: And I remind you againto recollect the beginning of our
discus-sionof what you two recommended: you said that the good
legislator shouldconstruct his entire legal code with a view to
war; for my part, I maintainedthat this was to order him to
establish his laws with an eye on only one virtueout of the four.
(Plato, Laws , tr. Saunders.)
.. Dont bring together everything proper to the hypothesis
, , .
ose who establish constitutions try to bring together every
single elementthat is proper to the hypothesis, but it is a mistake
to do this, as we said earlierin our discussion about how
constitutions are destroyed and preserved. (Poli-tics VI., -.)
To understand this remark, it is important to bear in mind that
Aristotle oers it
in the context of a discussion of democracy, which he regards as
a deviant (-
) form of constitution. A deviant constitution, such as
democracy or oli-
-
garchy, will be functional and stable only if it is moderate:
for many democratic-
seeming elements destroy democracies, and many
oligarchic-seeming elements
destroy oligarchies. It is like a hooked or snub nose, which may
be beautiful so
long as it deviates only moderately from the straight, but as
the curvature be-
comes more extreme, the nose will rst become unseemly (lose its
), and
nally stop appearing to be a nose at all (-). Likewise, a
constitution can
be adequate if it is somewhat democratic or oligarchic, but if
it goes too far it will
get worse, and nally stop being a constitution.
erefore, if you are establishing a constitution based on the
hypothesis of
freedom (or the hypothesis that it is best for the city to be
maximally free), you
would do well not to follow out every consequence of that
hypothesis, and not to
institute every possible measure that would promote freedom.
Otherwise your
eorts are likely to be self-defeating (you will not produce a
constitution at all, but
something analogous to the body part too distorted to qualify as
a nose), and at
best you will produce a very bad constitution.
.. External and internal evaluation
, , , , .
Concerning the constitution of the Spartans and of Crete, and
perhaps theother constitutions too, there are two things to
inquire: rst, whether anythinghas been legislated well or not well
in comparison to the best arrangement;second, whether anything has
been legislated contrary to the hypothesis andcharacter of the
constitution intended. (Politics II., .)
. Politics ., -. .
-
Aristotle announces two sorts of criticism to which a
constitution may be subject.
One is external, or absolute, and consists in seeing to what
extent the constitution
achieves or fails to achieve an unqualiedly optimal arrangement
of laws, oces,
and courts. e other is internal, and consists in seeing to what
extent the ele-
ments of the constitution harmonize with or contradict the
hypothesis and char-
acter () of the constitution envisaged by its legislators. I do
not know what
distinction (if any) Aristotle has in mind between a
constitutions hypothesis and
its character.
Of the criticisms that Aristotle proceeds to make, it is not
always obvious
which are external and which internal. However, we have seen a
clear case of ex-
ternal criticism in section ..: the Spartan constitution is
premised upon an in-
correct hypothesis, namely that the city is well-governed if and
only if it is well
prepared to conquer in war. ere are two places in which
Aristotle clearly marks
an internal criticism, by saying that some measure is
detrimental specically rela-
tive to a choice or intention of the legislator. First, the
women:
- . , , .
Further, the license of their women is detrimental both to the
purpose of theconstitution and to the happiness of the city. For
though the legislatorwants the entire city to be tough, he displays
his wish in relation to men,whereas he has completely neglected the
women. (Politics II., , .)
Women are not adequately trained and disciplined, and this is
not just bad in
itself, but undermines the aims and values to which Sparta
herself is committed. If
there is a dierence between hypothesis and character, I would
venture that this
case involves the constitutions hypothesis: the hypothesis (that
the city should be
primed for victory in war) requires that the citizens be tough,
and yet the laws al-
low half of the population to be undisciplined and
frivolous.
-
Second, the meals:
. , - , , . , .
Nor were things legislated well concerning the common messes
called phiditiaby the person who rst set them up. ey should have
been nanced fromcommon funds, as in Crete. Among the Spartans, each
person must con-tribute, even though some are very poor and cannot
lay out this expense, withthe result that the lawgiver achieves the
opposite of his purpose. For he wantsthe arrangement of the common
messes to be democratic, but they turn outnot at all democratic
when they are legislated in this way. (Politics II.,.)
Requiring every participant in the common meals to contribute
funds for
them undermines the democratic intent behind instituting such
meals in the rst
place. Of course the Spartan constitution as a whole was not
meant to be democ-
ratic, but it was meant to incorporate democratic as well as
oligarchic elements.
By botching one of its democratic elements, the legislator
failed to create or main-
tain the desired balance of democracy and oligarchy.
.. Defects in Carthages constitution
- , .
Most of the things one could criticize on account of its
deviations [from thebest] turn out to be common to all the
constitutions we have discussed. Of thethings to criticize in
relation to the hypothesis of aristocracy and polity, someincline
more towards the people, some towards oligarchy. (Politics
II.,.)
-
Aristotle distinguishes, in similar terms to what we saw above
(..), between ex-
ternal and internal criticism. From an external standpoint, we
can simply note all
errors and defects. On the other hand, we can criticize a
constitution based on
values internal to it, by pointing out ways in which its
arrangements conict with
the basic premise from which it was designed. at premise, as
weve seen, is re-
ferred to as the hypothesis of the constitution in question.
.. Education of reason and habit (missing the best
hypothesis)
. - - , .
It remains to consider whether children should be educated rst
by reason orby habits. For these must harmonize with each other in
the best way: for it ispossible for reason completely to miss the
best hypothesis, and for someone tobe similarly led by his habits.
(Politics VII., .)
As we know from the Nicomachean Ethics, complete human virtue
requires both
an excellent habituated condition of the non-rational part of
the soul and an ex-
cellent condition of the reasoning faculty. Here Aristotle briey
(and somewhat
cryptically) alludes to one of the reasons why both are needed.
If ones rational
faculty has not been properly educated, then he is liable to
reason from less than
optimal premises about what to do. (Given Aristotles model of
practical rea-
soning (cf. pp ), these will be premises about what ends should
be realized.
Perhaps, as Reeve (), p. thinks, Aristotle has in mind a single
premise
about the ultimate end, i.e., a view about what happiness is;
but it seems just as
likely that he envisages an ethical agent making a series of
judgments from day to
day as to what more immediate ends should be pursued in the
particular situa-
. I follow Newman () ad loc. in understanding to refer to
deviations from thebest arrangement (as at Pol ., ), rather than
deviations from the three correct forms ofconstitutions (tyranny
from kingship, oligarchy from aristocracy, democracy from
polity).
-
tions in which he nds himself.) If his habits are no good, then
no matter how
well he reasons, his behavior will be just as bad as if he had
no understanding.
.
So far, we have seen the word hypothesis used in Aristotle to
refer to various
sorts of things that may be assumed or posited. One such sort of
thing is a goal,
and Aristotle occasionally refers to goals as hypotheses.
However, we have seen no
reason to think that the word hypothesis means anything dierent
when refer-
ring to goals than when it refers to other assumed or posited
things, such as
premises in arguments. In both cases, it just means something
like assumption.
Now I will turn to the prepositional phrase , hypothetically
or
from a hypothesis. Given what hypothesis means in isolation, we
should ex-
pect this prepositional phrase to mean something like on an
assumption. On the
other hand, if hypothetically necessary means necessary for the
achievement of
an end, then hypothetically must be capable of meaning something
like relative
to an end. Apart from the disputed passages about necessity, are
there any places
where hypothetically appears to have an end-related meaning?
.. Syllogisms
e most numerous use of hypothetically is in the Prior Analytics,
in
connection with so-called hypothetical syllogisms. ese are
syllogisms such that
they or their conclusions are proven, argued, completed, or
agreed to from a hy-
pothesis. Aristotle oen refers to them as , syllo-
. See chapter ., note on page for references.
-
gisms from a hypothesis; I think we should understand a verb in
participle form
such as or , proven or argued.
Obviously, from a hypothesis has nothing to do with ends in this
connec-
tion. It just means something like based on an assumption, the
assumption being
perhaps that whatever entails something false is false, or that
whatever is true of
one is true of all the members of a given class.
.. Resulting
It makes no dierence that the impossibility resulted from a
hypothesis (
); for the hypothesis we took was possible, and when something
possi-
ble is assumed, nothing impossible should result from it
(Physics VII.,
). Here Aristotle is proving the impossibility of one
proposition, p, by as-
suming the truth of a second proposition, q, which is known to
be possible, and
showing that p & q entails something impossible. e method is
described at Top-
ics VII., -:
Examine not only whether something impossible follows
immediately fromthe thesis, but also whether it is possible for it
to obtain from a hypothesis (), as it does for those who say that
being empty is the same as beingfull of air. Clearly, if the air
goes out, the thing will be not less but more empty:so assuming
something, whether true or false (it makes no dierence), the oneis
removed and the other is not. Hence they are not the same.
. e terminology is introduced at Prior Analytics ., -, in the
follow-ing words: It is necessary that every demonstration and
every deduction should prove []either that something belongs or
that it does not, and this either universally or in part, and
furthereither probatively or hypothetically. Here hypothetically
modies prove.. See section .... , .. , , , , . ( ) , . .
-
In these passages again it is obvious that the meaning of from a
hypothesis
has nothing to do with ends. Aristotle is simply talking about
one thing being en-
tailed by another.
.. Knowing
In two passages in the Posterior Analytics, Aristotle speaks of
knowing some-
thing (eidenai, epistasthai) from a hypothesis. In one,
Aristotle rehearses an argu-
ment (which he rejects) that there is no knowledge; in the
other, he argues (in his
own voice) that there are no innite chains of predication in
demonstrative sci-
ence. Both arguments involve considering the case in which we
prove some-
thing, p, from premises that we do not know to be true.
Aristotle says that such a
proof does not yield knowledge of p without qualication, but
only knowledge of
p from some things or from a hypothesis ( , ; , ,
, ).
ough it isnt clear precisely how knowledge from a hypothesis
should be an-
alyzed, the general idea is clear enough: it consists, roughly,
in knowing (without
qualication) the truth of a conditional, i.e., that if one thing
(the hypothesis) ob-
tains, then another thing (the conclusion) obtains. e hypothesis
in question,
then, is a premise or conjunction of premises. Grasping a proof
of something, p,
from this premise or these premises, but not knowing that the
premises them-
selves are true, one knows p merely from a hypothesis (assuming,
of course, that
one does not know p in some other, independent way).
As before, it is obvious that the meaning of from a hypothesis
has nothing to
do with ends.
. Posterior Analytics ., -; ., -.. According to Barnes (), p. ,
Aristotles phrase is ambiguous between if is the case,then a knows
that P and a knows that if is the case then P. e latter is favoured
by the contextof the argument, and by Aristotles few words on the
hypothetical syllogism . But probably Aris-totle has not seen the
distinction.
-
.. Wishing
And he will wish good things [to his friend]: simply those that
are good sim-ply, and, for those that are good for that man, from a
hypothesis.
In the course of his discussion of friendship in Eudemian Ethics
book VII, Aristo-
tle considers the ways in which friendship is possible between a
decent man and a
bad man. As throughout the treatise, the text here is quite
damaged and it isnt
certain what Aristotle wrote. However, the particular clause Ive
quoted seems al-
right. Aristotle is distinguishing two ways in which the decent
man will wish his
not-so-virtuous friend well, based on a distinction he made
earlier, in VII.,
-, between two classes of good things (and, along the same
lines, two
classes of pleasant things): those that are simply good, or good
full stop, and those
that are good for someone but not good simply. To explain the
dierence, he says
that what is advantageous to a healthy body is simply good for a
body, whereas
what is advantageous only to a sick body, such as drugs or
incisions, is not simply
good for a body. He goes on to say, likewise in the case of soul
(): thus he
holds, quite generally, that what is good for a man in good
condition, i.e., a virtu-
ous and healthy man, is simply good, while what is good only for
a man in bad
condition is not simply good. It seems clear that the notion of
good for here
means what truly benets or improves the person, or what furthers
his achieve-
ment of truly good endswhether or not he himself wants it or
regards it as good,
and whether or not it furthers the achievement of any ends he
wishes to achieve.
. EE ., -. , [mss: ], .. See also EE VII., ; Pol VII., -; MM
II..... ree reasons: Aristotle has another, separate distinction
between what is good and what ap-pears good; my interpretation ts
the medical example; Aristotle says (-) that the decentman is
useful to the bad one relative to natural choice as opposed to
actually obtaining choice.
-
ings that are good for someone in bad condition are a sort of
corrective to a
bad situation.
Now, in friendship between good men, there is no dierence
between what is
good for the friend and what is good simply. But if your friend
is not altogether
virtuous, the two sorts of good will come apart. Aristotle says
that there is a quali-
cation to the way in which you will wish your friend the rst
kind of good: you
will wish him these not simply but from a hypothesis. What is
the nature of this
qualication?
Someone looking for an end-related meaning of hypothetically
might pro-
pose that wishing something hypothetically means wishing it for
the sake of some-
thing else, as opposed to wishing it for its own sake. In
support of this, one could
point to the end of the sentence whose beginning I quoted above:
[he will wish
for] these things [i.e. those that are good only for the friend]
for the sake of things
that are good simply, just like drinking medicine: he doesnt
wish it, but wishes it
for the sake of such-and-such. e thought would be that there is
a distinction
between wishing for something specically as a means, and wishing
for it, if not
necessarily as an ultimate end, at least not as a means to
anything in particular.
ings that are simply good, such as health and riches, are wished
for in the sec-
ond way, i.e., simply, whereas things that are good only for
someone in bad condi-
tion are always wished for hypothetically, i.e., as means to
some denite simply
good thing. On this proposed reading, the simply good thing
would be the hy-
pothesis relative to which the good-for-him thing is wished.
On the other hand, we could understand the qualication in a
dierent way,
one more in line with the uses of hypothesis we have seen
before. Namely, we
could understand wishing something hypothetically as wishing it
conditionally:
wishing it if , or given that . e thing wished for is only
desirable because
. EE VII., -. [mss. ] , {} , .
-
some undesirable condition now obtains, and so one doesnt wish
simply for the
medicine, or the punishment, or whatever, because one would
rather that the con-
dition making it desirable didnt obtain. Only holding xed that
so-and-so is in
such-and-such bad condition, do you wish for the thing in
question.
I dont know exactly how the notion of wishing something if q, or
wishing
something given that q, should be spelled out, but it is
evidently parallel to that of
knowing something hypothetically. I suggested that knowing q
hypothetically
might consist in knowing, for some p, that if p then q, while
knowing neither p
nor q. Wishing that q hypothetically could consist in wishing,
for some p, that if p
then q, but wishing neither that p nor that q. (I wish that if
his leg is gangrenous
then it is amputated, but I dont wish that his leg is gangrenous
or that his leg is
amputated.)
It is worth making a comparison to Socrates argument in Platos
Gorgias, c
., that tyrants and orators do not do what they wish. e argument
opens with
the claim that if someone does a for the sake of b, then he
wishes b, not a. Exam-
ples in the place of a are drinking medicines and sailing;
examples in the place
of b are health and wealth. us the sorts of things that
Aristotle will call simply
goodhealth and wealthare treated in the Gorgias as things wished
for their
own sake, or at any rate, things not wished for the sake of
something else.
Socrates goes on to introduce a partition of all things into
those that are good,
those that are bad, and those that are in between. ings in
between are ones that
sometimes partake of the good, sometimes of the bad, and
sometimes of neither.
Having secured an agreement that in-between things are always
done for the sake
of good things, or for the sake of the good, he infers:
en we do not wish to kill, or exile people from cities, or
conscate goods,simply just like that ( ), but rather if these
things are benecial we
. Sailing is described as a dicult and dangerous undertaking, .
Cf Dodds (), p. ad : e Greeks did not go on pleasure cruises, or
take sea voyages for their health; sail-ing was still a dangerous
business, as Demosthenes speeches on bottomry suciently show,
andoen highly uncomfortable; Hesiod thought it folly.
-
wish to do them, and if they are harmful we do not wish to do
them. For wewish good things, as you agree, not things that are
neither good nor bad, northings that are bad.
It isnt obvious that this is the right conclusion to draw, but
that is an issue we
neednt address here. What I want to take home is: (a) Socrates
distinction be-
tween what is good and what is in between, but partakes of the
good in a given sit-
uation, seems to match Aristotles distinction between what is
simply good and
what is not simply good but good for someone. (b) Socrates uses
the adverb sim-
ply () to modify wish (), as does Aristotle. (c) What simply
means in the Gorgias is without distinction, and the needed
distinction is ex-
pressed with if.
Something similar is going on in EE as in the Gorgias, and if EE
is an early
work we might expect it to formulate things in Academic ways. In
the Gorgias,
as in the EE passage, there is a distinction between wishing
something for the sake
of something else and wishing something in its own right. But
what is contrasted
to wishing something simply is wishing it if something is the
case, not wishing it for
the sake of something else. So understanding from a hypothesis
in EE to mean
conditionally or on an assumption makes Aristotles formulation
run parallel to
Platos. Combined with the fact that this reading best matches
Aristotles use of the
phrase elsewhere, this leaves us with good reason to accept the
reading.
.. Employment of virtue
( , ) , . , -, , ( -
. Gorgias c -, my translation.. It does so in other respects,
such as its use of the Academic term (employment) inplace of the
more typically Peripatetic (activity or actuality).
-
), - .
We say (and we have den